Just Leaving This Here...

It’s not my fault btw if people don’t have the power to understand math, statistics (as a field), and the fact that biased human opinion is meaningless. There’s a whole bunch of people here I envision being in Vegas laying money on a roulette table thinking the longer they play the better their odds get because “at some point odds say you have to win eventually.”
 
Here is the AI overview on Sagarin rankings.

Jeff Sagarin's Sagarin Rankings are based on two rating systems, each of which assigns a certain number of points to each team:
  • Elo chess
    This system is based on the Elo rating system used to rank chess players, and only considers wins and losses.
  • Predictor
    This system takes the margin of victory into account. The difference in rating scores between two teams is meant to predict the margin of victory for the stronger team.

In both systems, teams gain higher ratings by winning against stronger opponents, and home-venue advantage is also considered. In the Predictor system, a law of diminishing returns is applied to the margin of victory. For example, a team that wins by 7–6 is rewarded less than a team that wins by 21–7, but a team that wins by 35–0 receives similar ratings to a team that wins by 70–0.

I will add that Sagarin is not based on any "in-game" performance metrics, such as yards or completions. It's just W/L and score of a team, its opponents, and its opponents opponents.
 
That was two months ago...with McNamara at QB. The question is who is a better team right now. If they played again this Saturday, you betting on Iowa State?
That would be strictly a matter of subjective opinion, whether it be by me, you, or a programmed computer. On the other hand, an objective game has already been played between the two teams, with an objective timekeeper and referees. The result of the game is a proven fact. Until next season, no one has any basis to say Iowa is a better team than Iowa State. You can say it, but it has no basis in fact or logic, it's just opinion.
 
Where I struggle is the suggestion that it is all or nothing. I am not one to normally say that because a robot says it, it must be correct. I also recognize human polling bias. Both have flaws. Here, Sagarin's system has an obvious flaw if it ranks Iowa ahead of Iowa State right now. But, the truth can be somewhere in the middle.

As far as rankings go, there is a flaw in Gry's logic. Predicting who will beat another team is not the same thing as ranking a team's performance based upon the body of work to date. Just because one team is ranked ahead of another does not mean that they will be predicted to win the game. Lots of variables go into that.

Predicting a winner is uncertain and based upon guesswork based upon events and actions that have yet to occur as extrapolated from what has occurred to date.

Ranking is judging the worthiness of a team based upon events that have already occurred in the past (ie which games they have won and lost). Indiana is a perfect example. Right now, they should be ranked based upon what they have done to date, not guessing how they might fare in the tougher games coming up. Once those games are decided, then you weave those variables into your rankings.
 
O
As far as rankings go, there is a flaw in Gry's logic. Predicting who will beat another team is not the same thing as ranking a team's performance based upon the body of work to date. Just because one team is ranked ahead of another does not mean that they will be predicted to win the game. Lots of variables go into that.
I've not said there's no flaw in the logic. I'm saying the flaws in the logic of trusting an objective ranking formula are much less than the flaws in trusting votes cast by a bunch of extremely biased reporters who've never played nor coached football.

When it comes to Sagarin, it's not just predicting wins/losses. The wins/losses predictions with that system have been pretty rigorously shown that his particular method is WAY more accurate than any other system or voting poll out there. The win/loss prediction is just a way of measuring the validity of the system. That's why Sagarin includes that metric, because he can offer as many rankings as he wants but how do you measure the validity otherwise? If there's a better method I'm sure he'd be all ears...he's got a math degree from MIT; I'm trusting his intuition and results more than anyone else here including myself. Look up the study done on his rankings as a predictor of success against the two main polls. It's more than significant enough (strikingly so, in fact) to show you that the two polls are wildly off-base. The study used the assumption that a team ranked 10th in either system should beat teams ranked lower than them and so on. Totally valid and beautifully simple. Is it a perfect predictor? Obviously not. But it's miles better ranking than anything else out there. Any method of ranking college football teams is going to be crippled by horribly small sample sizes.

And when you plug Iowa/Iowa State into that system equally and without any outside human bias, Iowa comes out one spot ahead of ISU.

What I think one of the main takeaways of my original post should be is that Iowa receiving 2 votes and not appearing in the AP top 25 is a whole lot dumber than Iowa being ranked 16th in the country when considering their opponents to date. In other words, my screenshot at the top of the page shows just how stupid and irrelevant (to pure rankings, not opinion) the voting polls are.
 
Last edited:
It’s not my fault btw if people don’t have the power to understand math, statistics (as a field), and the fact that biased human opinion is meaningless. There’s a whole bunch of people here I envision being in Vegas laying money on a roulette table thinking the longer they play the better their odds get because “at some point odds say you have to win eventually.”

I am a big numbers guy.
I'm a huge fan of numbers. I like talk data all the time and fashion myself as an amateur data analyst. Sometimes what I post is based on fallible data purporting to support a flimsy arguments. One of my favorite TikToks at the moment is this guy that does "things that shouldn't match, but do".
And he'll show graphs demonstrating near perfect correlations of things like "the popularity of the first name Nelson matches cottage cheese consumption" or "the number of plumbers in Texas matches google searches for flights to Antartica".

That being said, data driven predictions are fallible as well. Less so, no doubt, than coaches or sports writers.

I do think "today's" Iowa team beats the ISU team that beat the Hawks. I think they beat the ISU team of today as well, at home. For sure. On the road? A bit tougher call. But that's just what I think. But, I also bristle at the idea of data somehow proving that.

In the end, this is why I CANNOT stand the nature of this playoff system. If we're going to do this, it's very easy. Figure out which conferences you want to include. Let the conference determine a champion. The champions go to the playoffs. I'm fine with rankings determining seedings, and can live with them determining any necessary 'byes'. My family is split over this. Some think that a Big 12 (or even lesser conference) matchup against an Oregon is "bad football" to watch on TV. We're not trying to create "good TV". The point is determining a champion. Count me as old school. I think the wild cards are a bad introduction to baseball. Win your damn division. Win your damn conference. Especially in a sport like college football where time/scheduling limitations exist.
 
I am a big numbers guy.
I'm a huge fan of numbers. I like talk data all the time and fashion myself as an amateur data analyst. Sometimes what I post is based on fallible data purporting to support a flimsy arguments. One of my favorite TikToks at the moment is this guy that does "things that shouldn't match, but do".
And he'll show graphs demonstrating near perfect correlations of things like "the popularity of the first name Nelson matches cottage cheese consumption" or "the number of plumbers in Texas matches google searches for flights to Antartica".

That being said, data driven predictions are fallible as well. Less so, no doubt, than coaches or sports writers.

I do think "today's" Iowa team beats the ISU team that beat the Hawks. I think they beat the ISU team of today as well, at home. For sure. On the road? A bit tougher call. But that's just what I think. But, I also bristle at the idea of data somehow proving that.

In the end, this is why I CANNOT stand the nature of this playoff system. If we're going to do this, it's very easy. Figure out which conferences you want to include. Let the conference determine a champion. The champions go to the playoffs. I'm fine with rankings determining seedings, and can live with them determining any necessary 'byes'. My family is split over this. Some think that a Big 12 (or even lesser conference) matchup against an Oregon is "bad football" to watch on TV. We're not trying to create "good TV". The point is determining a champion. Count me as old school. I think the wild cards are a bad introduction to baseball. Win your damn division. Win your damn conference. Especially in a sport like college football where time/scheduling limitations exist.
This would be similar to the Mens NCAA BB tourney up until the expansion of the tournament field.
For those of you too young to remember, for years the only way to qualify for the NCAA was to win your conference. There were handful of slots for independent teams. But if you were in a conference you had to win it to be in. 2nd place was a trip to the NIT (which often had a large number of really good teams).
As an example, here is a link to the 1970 field.


There are pros & cons, but I think most would agree the expansion of the field greatly increased interest in the tourney and college BB in general.
 
Top